Snapshot spatial–temporal compressive imaging

Abstract

We build a compressive imaging system that can reconstruct high-speed videos of two fields-of-view by capturing a single frame. A single digital micromirror device is employed to implement high-speed modulation simultaneously on both scenes in a manner that does not compromise spatial resolution. We utilize deep denoising priors in the plug-and-play framework for efficient reconstruction. Two spatial-temporal data cubes, each with the size $ 650 \times 650 \times 20 $650×650×20, are reconstructed from a single two-dimensional spatial measurement.

Topics & Concepts